This study was designed to assess the efficacy and safety of substituting cisplatin with oxaliplatin in the DHAP (dexamethasone, cytarabine and cisplatin) regimen for patients with relapsed or refractory non-Hodgkin's lymphoma. Twenty-four evaluable patients with intermediate or high-grade non-Hodgkin's lymphoma were treated at 3-weekly intervals with oxaliplatin (130 mg/m2, d 1), cytarabine (2 g/m2 for two doses, d 2) and dexamethasone (40 mg, d 1-4). The median age of the patients was 58 (range 18-70). Histological subtypes were diffuse large B cell, 20; mantle cell, two; anaplastic large cell, one; and peripheral T cell, one. The overall objective response rate (RR) was 50% [95% confidence interval (CI) = 29-71%] including four complete responses and eight partial responses. RR for those patients treated at first relapse was higher than those treated at second and subsequent relapse (77% versus 29%). Grade 3 and 4 toxicity was mainly haematological: anaemia 17%, neutropenia 75% and thrombocytopenia 75%. No grade 4 non-haematological toxicity was reported. No significant renal and neurotoxicity was demonstrated. Median survival was 10.6 months. Probabilities of 1-year progression-free survival and overall survival were 47% (95% CI = 26-66%) and 50% (95% CI = 23-72%) respectively. In conclusion, dexamethasone, cytarabine and oxaliplatin (DHAX) is a novel combination in salvage therapy for relapsed or refractory non-Hodgkin's lymphoma. It has clinically significant activity with an acceptable toxicity profile. Lack of renal toxicity makes DHAX an attractive cytoreductive regimen before high-dose chemotherapy.